blofin-proxy

A local WS-fed caching reverse proxy for the BloFin API, built for running multiple freqtrade bots against BloFin without getting the IP rate-limited.

Each bot independently polling REST (market/candles, mark-price-candles, tickers, books, ...) per pair per few seconds multiplies into a ban: N bots × M pairs × K timeframes. This proxy collapses that traffic to:

• one WebSocket connection (via ccxt.pro.blofin) feeding a live candle and ticker cache, and
• single-flight + TTL-cached upstream REST for everything else, all gated by one shared token-bucket rate limiter (default 8 req / 2 s).

Private/signed endpoints (orders, balance, ...) are forwarded verbatim — BloFin signs path+method+timestamp+body, not the Host header, so trading keeps working through the proxy unchanged.

Install

pip install -r requirements.txt   # aiohttp, ccxt>=4.5

Run

cd blofin-proxy
python -m blofin_proxy --port 8095 --host 127.0.0.1

Options: --upstream (default https://openapi.blofin.com), --rate (default 8), --rate-window (default 2.0 s), --log-level.

Sanity checks:

curl "http://127.0.0.1:8095/health"
curl "http://127.0.0.1:8095/api/v1/market/candles?instId=BTC-USDT&bar=5m&limit=5"

/health reports uptime, WS watcher status, cache hit/miss stats, and the total upstream request counter (proof that bot traffic collapsed).

Run as a service

blofin-proxy.service is a systemd unit for the common case (venv at /home/ron/.venv, checkout at /opt/freqtrade/blofin-proxy). Edit User, WorkingDirectory and ExecStart to match your paths, then:

sudo cp blofin-proxy.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now blofin-proxy
systemctl status blofin-proxy
journalctl -u blofin-proxy -f

Start it before the bots — freqtrade fails its market-data calls if the proxy is not listening.

Point freqtrade at the proxy

In each bot's config:

"exchange": {
  "name": "blofin",
  "ccxt_config": {
    "enableRateLimit": false,
    "urls": {"api": {"rest": "http://127.0.0.1:8095"}}
  },
  "ccxt_async_config": {
    "enableRateLimit": false,
    "urls": {"api": {"rest": "http://127.0.0.1:8095"}}
  }
}

enableRateLimit: false is safe because the proxy is now the single gatekeeper for upstream REST traffic.

How endpoints are served

Endpoint (/api/v1/market/...)	Strategy
candles	WS-fed candle book; cold keys seed from REST (single-flight) and start a watcher
mark-price-candles, index-candles	TTL = max(5 s, bar/4) + single-flight
tickers	WS-fed when available; TTL 5 s cached fallback
books	TTL 1 s + single-flight
trades, mark-price	TTL 2 s + single-flight
instruments, position-tiers	TTL 300 s
funding-rate, funding-rate-history	TTL 60 s
everything else (any method)	forwarded verbatim, streamed back through the limiter

Development

# from the repo root
python -m pytest -q

Layout: blofin_proxy/limiter.py (token bucket), cache.py (TTLCache + CandleBook), upstream.py (forwarding client), server.py (routes + catch-all), feed.py (ccxt.pro WS watcher manager), __main__.py (CLI).

Referral

If you like this tool signup at BloFin with my referral: https://blofin.com/register?referral_code=HYXOVL

License

MIT — see LICENSE.